Impact of assimilating the VIIRS-based CrIS cloud-cleared radiances on hurricane forecasts

Jun Li@, Pei Wang@, Jinlong Li@, Zhenglong Li@, Jung-Rim Lee&, Agnes Lim@, Timothy J. Schmit#, and Mitch Goldberg%,

@CIMSS, University of Wisconsin-Madison%JPSS Program Office, NESDIS/NOAA

&Korea Meteorological Administration#SaTellite Applications and Research (STAR), NESDIS/NOAA

The 21th International TOVS Conference (ITSC-21)29 November – 05 December 2017, Darmstadt, Germany

Acknowledgement: This research is partly supported by JPSS Proving Ground program

1

Outlines

Motivation and current status on using imager to assist handling clouds for IR sounder radiance assimilation;VIIRS-based CrIS cloud-cleared radiances (CCRs);Impact from CCRs and GOES-16 moisture on

recently hurricanes;Summary and future work.

2

AIRS, IASI, CrIS, HIRAS (LEO sounder)GIIRS (GEO sounder)

MODIS, AVHRR, VIIRS, MERSI (LEO Imager)AGRI (GEO Imager)

1. Both instruments (sounder and imager) have good signal-to-noise ratio, have overlap spectral coverages;2. They are comparable through convolving sounder to imager spectrally and averaging imager to sounder spatially.

Using high resolution imager measurements to assist hyperspectral IR sounder radiance assimilation

3

Using imager for assisting IR sounder radiance assimilation:1. IR sounder sub-pixel cloud-detection and QC for radiance assimilation;2. Cloud-cleared radiances (CCRs) in cloudy skies for assimilation.

Current status on using imager for sounder radiance assimilation

IR sounder sub-pixel cloud characterization has been demonstrated and recommended by ITWG to operational centers at ITSC20;

Using collocated imager data (IR band radiances and cloud mask) to derive the sounder cloud-cleared radiancesAIRS/MODIS demonstrated (Wang et al. 2016);Algorithm implemented for CrIS/VIIRS, demonstrated with

CIMSS SDAT for recent hurricanes;VIIRS-base CrIS CCRs tested in GFS by EMC (Dr. Collard and Dr.

Liu) (9p.09 - Liu et al.);VIIRS-based CrIS CCRs tested in RAP by ESRL;Test by NRL planned (Dr. Ruston). 4

SDAT(http://cimss.ssec.

wisc.edu/sdat)

Research testbed for improving the

assimilation of JPSS/GOE-R data

(Sounder, ABI, radiances, TPW,

AMVs, Clouds, etc.)

R2O Research

Operational NWP models:

EMC GFS

EMC HWRF

ESRL RAP

End Users (NHC and Local Forecasters)

5

CIMSS near real-time Satellite Data Assimilation for Tropical storms forecasts (SDAT) (http://cimss.ssec.wisc.edu/sdat)

WRF/GSI at CIMSS

TC Forecast Improvement

Applications

Refining the Operational Path

CIMSS SDAT 2017090806 00-72 hours forecasted IR-WV ABI Imager - IRMA

6

CrIS/VIIRS cloud clearing for CrIS radiance assimilation

Clear Cloud

The CC method (Li et al., 2005);

VIIRS cloud mask identifies partially cloudy FOVs (black circle);

VIIRS radiances help quality control cloud cleared CrIS radiances;

Only three VIIRS bands (4.05, 10.763, and 12.013 um) used (overlapped with CrIS);

Cloud cleared radiances very close to VIIRS clear sky radiances;

12.5 % of partially cloudy FOVs are successfully cloud cleared for Hurricane Sandy (2012) case.

Cloud impact removed!

7

CrIS clear radiances: (clear FOVs + radiances not affected by clouds)

Example of 1 day CrIS coverage 8

CrIS CCRs:(clear + CC coverage)

8

Analyzing 120hr forecasts from 18 UTC 09-30 2015

CrIS original radiances assimilated at 18 UTC 09-30 (channel 130)

CrIS CCRs assimilated at 18 UTC 09-30 (channel 130)

How do CrISCCRs improve tropical cyclone forecast?

9 9

BT (K) BT (K)

CrIS Original CrIS CC

120 hr forecasting from 09-30 18z 500hPa

Hurricane Joaquin is merged with the low pressure center over CONUS with assimilation of CrIS radiances (left) Hurricane Joaquin is separated from the low pressure center over CONUS with assimilation of CrIS CCRs

(right), which is verified with GOES Imager. 10

BT (K)

Experiments on Hurricane Harvey (2017)WRF-ARW v3.6.1: 12 km horizontal resolution (400*300) , 52 vertical layers from surface to 10hPa

GSI v3.3: 3D-Var Data Assimilation Method• NAM background error covariance matrix• Conventional Data (GTS)• AMUS-A radiances onboard NOAA-15, NOAA-18, NOAA-19,

and Metop-A• IASI onboard Metop-A and Metop-B• ATMS onboard Suomi-NPP• CrIS radiances onboard Suomi-NPP• Updated bias correction for each cycling, enhanced bias

correction method in GSI• Background and initial conditions: NCEP FNL (BG: GFS).

72h

72h6h

… …

23 00z

23 06z

Forecast

Data

6h

Forecast

Data

Hurricane Harvey (2017)• Assimilation : Aug 23 00z to Aug 25 18z, 2017• Forecasts: Aug 23 12z to Aug 28 18z, 2017• Assimilation every 6 hour, 10 groups in statistics

Data:Conv from GTS;POES: AMSU-A, IASI, ATMS and CrIS;CCRs: CrIS cloud-cleared radiances (CCRs) in cloudy skies;GOES-16: Three layered precipitable water (LPW) from ABI at: 0.3 - 0.7, 0.7 - 0.9, and 0.9 – 1.0 in sigma level.ExperimentsCNTRL: Conv+AMSUA+IASI+ATMS+CrIS (Conv + POES)CNTRL+CCRs: adding radiances in cloudy skiesCNTRL+LPW: adding GOES-16 moisture information

Model domain

11

L-LPW: 900 hPa - SFC

2017-8-24 06zGOES-16 Observations O-B

M-LPW: 700 – 900 hPa

H-LPW: 300 – 700 hPa

(mm) (mm)

12

Statistics (RMSE) from the experiments

Forecast Time (hour)

13

200 – 1000 hPa 300 – 1000 hPa

200 – 1000 hPa

200 – 1000 hPa 200 – 1000 hPa

25 – 30 km improvement over control

14

BT of GOES 16 6.95 µm band from 2017-8-25 06z to 8-28 06zObservations

CNTRL

CNTRL+CCRs CNTRL+LPW

Better moisture distribution 15

ETS scores for 06 – 48 hour forecasts10 mm precipitation (and >)

20 mm precipitation (and >)

1 mm precipitation (and >)

15 mm precipitation (and >)

16

Obs HRRR Exp

25 06z

25 12z

Heavy Precipitation (> 100 mm/6hr) over land

6 h forecasts

25 18z

Two RainbandTwo Rainband18 h forecasts

6-hour cumulative precipitation forecasts started at 00 UTC on 25 August 2017

12 h forecasts

17

0

50

100

150

200

250

300

Conv C_LPW C_LPW_RAD

Mean RMSE of track (Km)

0

50

100

150

200

250

300

350

400

450

0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96 102

108

114

120

Track forecast RMSE (km)

Conv

C_LPW

C_LPW_RAD_CCR

2017090512 – 2017091018, 5-day Irma forecasts updated 6-hourly

Assimilate Conventional data onlyConventional; LEO: IASI (Metop-A/-B), AMSUA (Metop-A/-B, NOAA-15/-18/-19), CrIS, CCRs, ATMS; GEO: LPW (GOES-16)

18

Operational models compare to the best-track estimate(2017090512 - 2017091018)

19

Summary and future workAssimilating CrIS CCRs show positive impact on

hurricane track forecasts, could be an alternative of radiance assimilation in cloudy skies;

QC is important, since atmosphere might be inhomogeneous within IR sounder sub-pixel in cloudy condition;

Future work will focus on full spectral resolution CrIS from NOAA-20, improve QC on CrIS CCRs, collaborate with users on more experiments in NOAA and other models for potential operational application.

20